For the problem of anti-background interference of weak-small targets in infrared images, target extraction and texture detail processing are key tasks in the image fusion algorithm. The single-band infrared data can not fully reflect image details and contour information. There are texture differences in different bands of data, which makes it difficult to recognize targets. Therefore, it is necessary to fuse dual-band data to identify weak-small targets clearly. To solve these question, in this paper, we propose an effective image fusion framework using Latent Low-Rank Representation (LatLRR) and Discrete Wavelet Transform (DWT). Firstly, all source images are trained as L matrix by LatLRR which is used to extract salient features. And the original images are decomposed into high frequency and low frequency by DWT. Then high frequency parts are fused by maximum absolute value and low frequency parts are fused by weighted-average. On this basis, the training matrix L and high frequency fusion parts are used for contrast modulation fusion. Finally, the fused image is reconstructed by combining the contour parts and feature parts. The experimental results demonstrate that our proposed method achieves state-of-the-art performance in objective and subjective assessment.

Under night vision conditions, visible-spectrum images often fail to capture background details. Conventional visible and infrared fusion methods generally overlay thermal signatures without preserving latent features in low-visibility regions. This paper proposes a novel deep learning-based fusion algorithm to enhance visual perception in night driving scenarios. Firstly, a local adaptive enhancement algorithm corrects underexposed and overexposed regions in visible images, thereby preventing oversaturation during brightness adjustment. Secondly, ResNet152 extracts hierarchical feature maps from enhanced visible and infrared inputs. Max pooling and average pooling operations preserve critical features and distinct information across these feature maps. Finally, Linear Discriminant Analysis (LDA) reduces dimensionality and decorrelates features. We reconstruct the fused image by the weighted integration of the source images. The experimental results on benchmark datasets show that our approach outperforms state-of-the-art methods in both objective metrics and subjective visual assessments.